Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
ORIGINAL RESEARCH PAPER
4D seismic time-lapse monitoring of an active cold vent, northernCascadia margin
Michael Riedel
Received: 28 June 2007 / Accepted: 3 December 2007 / Published online: 5 January 2008
� Springer Science+Business Media B.V. 2007
Abstract Two single-channel seismic (SCS) data sets
collected in 2000 and 2005 were used for a four-dimen-
sional (4D) time-lapse analysis of an active cold vent
(Bullseye Vent). The data set acquired in 2000 serves as a
reference in the applied processing sequence. The 4D
processing sequence utilizes time- and phase-matching,
gain adjustments and shaping filters to transform the 2005
data set so that it is most comparable to the conditions
under which the 2000 data were acquired. The cold vent
is characterized by seismic blanking, which is a result
of the presence of gas hydrate in the subsurface either
within coarser-grained turbidite sands or in fractures, as
well as free gas trapped in these fracture systems. The
area of blanking was defined using the seismic attributes
instantaneous amplitude and similarity. Several areas
were identified where blanking was reduced in 2005
relative to 2000. But most of the centre of Bullseye Vent
and the area around it were seen to be characterized by
intensified blanking in 2005. Tracing these areas of
intensified blanking through the three-dimensional (3D)
seismic volume defined several apparent new flow path-
ways that were not seen in the 2000 data, which are
interpreted as newly generated fractures/faults for upward
fluid migration. Intensified blanking is interpreted as a
result of new formation of gas hydrate in the subsurface
along new fracture pathways. Areas with reduced
blanking may be zones where formerly plugged fractures
that had trapped some free gas may have been opened and
free gas was liberated.
Keywords 4D seismic time-lapse imaging �Seismic processing � Gas hydrate � Cold vent �Fracture systems
Abbreviations
IODP Integrated Ocean Drilling Program
LWD Logging-while-drilling
NEPTUNE Northeast Pacific time-series undersea
networked experiments
RMS Root-mean square
SCS Single channel seismic
TWT Two-way travel time
2D Two-dimensional
3D Three-dimensional
4D Four-dimensional
Introduction
Cold vents are commonly observed on active margin set-
tings where they contribute significantly to the fluid flow
within the accretionary prism. Such active cold vents were
observed for example on the Aleutian margin (Suess et al.
1998), at Hydrate Ridge on the Cascadia margin offshore
Oregon (e.g., Suess et al. 1999, 2001), offshore Vancouver
Island (Riedel et al. 2002, 2006a), at the Nankai Trough
offshore Japan (e.g., Kobayashi 2002 and reference
therein), and along the Makran margin (e.g., von Rad et al.
2000). They are also seen in passive margin settings such
as offshore Nova Scotia (Shimeld et al. 2004), at the Blake
Ridge (Paull et al. 1996) and in the Gulf of Mexico (e.g.,
Sassen et al. 2001 and references therein). Cold vents
are commonly associated with seafloor chemosynthetic
communities (clam colonies and bacterial mats) and with
M. Riedel (&)
Department of Earth and Planetary Sciences, McGill University,
3450 University Street, H3A2A7 Montreal, QC, Canada
e-mail: [email protected]
123
Mar Geophys Res (2007) 28:355–371
DOI 10.1007/s11001-007-9037-2
wide-spread carbonate formations. Often an association
with near-seafloor gas hydrate is found.
Active cold vents are areas where changes of the sea-
floor and subsurface conditions can change rapidly.
Probably the best studied site is Hydrate Ridge offshore
Oregon where rapid changes of gas venting were observed
(e.g., Tyron et al. 1999; Suess et al. 1999; Collier et al.
1999). Typically active gas venting has been observed at
several individual smaller vent outlets that can switch in
location rapidly within days or weeks (G. Bohrmann,
personal communication). The reason for the change in the
vent outlet location and activity can only speculated upon
as only few long-term monitoring data sets are presently
available. Changes in the subsurface, potentially the cause
for observable changes at the seafloor, have not been
monitored systematically.
The area of this study is located on the northern
Cascadia margin offshore Vancouver Island (Fig. 1) at
Integrated Ocean Drilling Program (IODP) Site U1328.
The vent field is characterized by several seismic blank
zones varying in size from a few tens of meters to several
hundred meters in diameter. Bullseye Vent is the largest
and most prominent vent in this vent field and has been a
site of intense studies over the past years. Studies include
numerous two-dimensional (2D) and three-dimensional
(3D) single and multichannel seismic surveys (Riedel et al.
2002), Ocean-Bottom-Seismometer surveys (Hobro et al.
2005; Spence et al. 1995; Zykov and Chapman 2004),
piston coring with physical property measurements
(Novosel 2002; Novosel et al. 2005) and geochemical
analyses (Solem et al. 2002; Pohlman et al. 2003; Riedel
et al. 2006a), heat flow measurements (Riedel et al.
2006a), bottom-video observations (Riedel 2001; Riedel
et al. 2006a), controlled-source electromagnetic imaging
(Schwalenberg et al. 2005) and seafloor compliance mea-
surements (Willoughby et al. 2005). Bullseye Vent was
also a target of the most recent IODP Expedition 311
during which five holes were drilled at Site U1328 along a
100 m section through the centre of the vent (Riedel et al.
2006b).
This paper presents an attempt at four-dimensional (4D)
seismic time-lapse imaging of an active cold vent. This
paper has two goals: (a) to demonstrate the technique of
how to process challenging unconventional single-channel
seismic (SCS) data sets for 4D time-lapse monitoring, and
(b) to demonstrate that changes did indeed occur within the
cold vent over the period of 5 years (2000–2005) and can
be mapped accurately.
Detecting temporal changes in the subsurface using
geophysical imaging techniques is by now routine in the
oil-and-gas industry (e.g., Lumley 2001 and references
therein). A standard seismic processing technique in
4D time-lapse monitoring is known as cross-equalization.
The main purpose of 4D seismic cross-equalization is to
reduce differences in areas with no expected changes and
to optimize 4D difference anomalies. Examples of using
seismic cross-equalization in 4D imaging can for example
be found in Eastwood et al. (1998), Harris and Henry
(1998), Naess (2006), and Gan et al. (2004).
The cross-equalization technique was adopted in this
study to compare two SCS data sets acquired over Bullseye
Vent where the location and nature of subsurface changes
(if at all present) are not previously known. The two data
sets were collected with similar acquisition parameters and
geometries, but substantial systematic differences were
observed between them. An enhanced seismic imaging
Fig. 1 (a) Map of general location of survey area on the northern
Cascadia margin. The zone of regional gas hydrate occurrence is
shown in gray inferred from the occurrence of bottom-simulating
reflections in regional seismic data. (b) Detailed map of the area of
the cold vent field near Ocean Drilling Program Site 889, IODP Site
U1327 and U1328. (c) Detailed outline of Bullseye Vent and location
of piston cores and IODP Site U1328 drill holes. The shaded outline is
the area of active venting, characterized by seismic blanking
356 Mar Geophys Res (2007) 28:355–371
123
technique for defining areas affected by fluid flow and/or
enhanced gas hydrate concentrations by using 3D geo-
metrical seismic attributes such as similarity (also referred
to as seismic coherency) is introduced. A potential limi-
tation of this time-lapse analysis of SCS data may be that it
is more sensitive to noise, which is not attenuated with
stacking. High frequency signals yield better resolution for
small changes of reflectors, but the use of high frequencies
demands an efficient noise removal filter. In this case study
where relatively high frequency signals were used, but
noise was difficult to attenuate, we introduce seismic
attributes that may help overcome this issue (i.e., similar-
ity) and limit our interpretation to only large-scale
variations that can be observed consistently across the 3D
seismic data.
The results of this 4D time-lapse imaging showed that
even relatively low-quality 3D seismic reflection data can
be used to detect subsurface changes if the seismic data is
processed appropriately. However, significantly better
results can be expected if for example the seismic receivers
are permanently placed on the ocean floor, which not only
reduces navigation uncertainties, but also increases the
signal-to-noise ratio. The study presented here shows that
subsurface changes in the cold vent are to be expected to be
detectable over timescales of a few years or less with
repeat-seismic surveys, and can be imaged using advanced
4D time-lapse processing and imaging techniques. This
result has important implications for the application of 4D
seismic imaging using the NEPTUNE cable observatory
that is currently implemented on the Cascadia margin.
Summary of Bullseye Vent observations
Bullseye Vent is part of a larger cold vent field that covers
an area of about 2 9 4 km on the mid slope of the northern
Cascadia margin (Fig. 1). Four main vents were identified
in this vent field and vary in size from a few tens of meters
to several hundreds of meters in diameter. All vents are
seismically characterized by reduced reflection amplitude
or blanking.
Massive pieces of gas hydrate were recovered by piston
coring (maximum 8 m penetration) at Bullseye Vent as
near as 0.5 m below the seafloor (Riedel et al. 2006a).
Logging-while-drilling (LWD) measurements at IODP Site
U1328 showed a zone of very large electrical resistivity of
up to 30 X m within the top 40–50 m below seafloor
(mbsf) in the centre of the vent (Riedel et al. 2006b). IODP
coring at this site also recovered several massive chunks of
gas hydrate within the top 50 mbsf, but gas hydrate was
only sporadically recovered at greater depth. LWD data at
IODP Site U1328 imaged several steeply dipping fractures
of very high electrical resistivity representing high gas
hydrate concentration within the fracture (Riedel et al.
2006b).
The controlled-source electromagnetic data also showed
large resistivity anomalies across all vents, with electrical
resistivities as high as 6 X m, which is around five times
the assumed regional no-gas hydrate background, sug-
gesting the presence of high gas hydrate concentrations at
those vents (Schwalenberg et al. 2005). Additional con-
straints on the subsurface plumbing were derived from
several seafloor compliance measurements at and around
Bullseye Vent. The compliance data show an overall
increase in shear-modulus (and hence shear-wave velocity)
below the vent within the depth range of the gas hydrate
stability zone, whereas no such pronounced shear-modulus
anomaly was seen away from the vent (Willoughby et al.
2005). The increase in shear-wave velocity is interpreted as
additional evidence for the presence of gas hydrate within
the cold vent. However, the compliance data do not provide
fine resolution of the depth distribution of gas hydrate.
Repeated seafloor video observations in 2000 and 2001
showed that the seafloor around Bullseye Vent is charac-
terized by wide-spread carbonate platforms and nodules, as
well as isolated chemosynthetic communities (Riedel 2001;
Riedel et al. 2006a; Beaudet et al. 2001). Water samples
collected over an active chemosynthetic community in
2001 showed evidence of methane released into the water
column up to a height of 200 m above the seafloor (Solem
et al. 2002). Surveys conducted in September 2006 using a
12 kHz echosounder, mapped extensive gas discharge
across Bullseye Vent where the plume structures rise to
*500 m below sea surface.
Measurements of physical properties of the top 8 m of
sediments within and around Bullseye Vent showed further
evidence of a long-time gas flux history. Magnetic sus-
ceptibility is reduced by several orders of magnitude inside
Bullseye Vent as result of chemical reduction of magnetic
minerals to pyrite through enhanced methane flux (Novosel
et al. 2005). Geochemical pore-water analyses further
demonstrated that the sulfate–methane interface is much
shallower inside Bullseye Vent than outside. Near the area
of the active chemosynthetic community the SMI is right at
the seafloor (Pohlman et al. 2003).
A model for seismic blanking
The nature of the seismic blanking has been highly debated
in the literature (Lee and Dillon 2001; Riedel 2001; Riedel
et al. 2002; Wood et al. 2002; Zuhlsdorff and Spiess 2004).
A discussion and comparison of these different models
was given by Riedel et al. (2006a). Their integrated model
appears most closely to the combined geophysical imaging
results and the IODP drilling. This model describes
Mar Geophys Res (2007) 28:355–371 357
123
Bullseye Vent as a complex subsurface network of frac-
tures partially filled with gas hydrate, feeding methane to
the near-surface where massive gas hydrate can form as
methane solubility drastically decreases. Seismic blanking
(seen as vertical wipe-outs) is therein a result of a combi-
nation of transmission losses at the massive gas hydrate cap
and subsurface gas hydrate accumulations (as found in
fractures). Laterally extensive blanking can occur by the
preferential gas hydrate accumulation in the coarser-
grained turbidite layers following the model by Lee and
Dillon (2001). The frequency dependent nature of the
blanking described by Riedel et al. (2002) is related to the
magnitude of seismic waves being affected by these het-
erogeneities in the subsurface. Lower seismic frequencies
(and thus larger wavelengths) are able to penetrate deeper
as they are less affected by small-scale heterogeneities (and
thus will be less attenuated) than higher frequencies with
smaller wavelengths.
4D time-lapse processing sequence
In this section the processing sequence used to match the
two data sets is described in detail. Although 4D seismic
imaging is by now established in the oil-and-gas industry,
it has not yet widely been used in academic science.
Furthermore, the unusual nature of the data sets (with
respect to standard oil-and-gas industry) and their appli-
cation to a cold vent environment, both require additional
care to ensure that no artifacts are introduced resulting in
false interpretation of the data.
The seismic processing steps for the complete 4D seis-
mic analysis and comparison are summarized in a flow
chart in Fig. 2. The two SCS data sets were not migrated
for this analysis because only sparse reliable velocity
information is available and migration with inaccurate
velocities would result in degraded imaging results. Prior
to any 4D processing and interpretation, both data sets
were simply band-pass filtered to remove some of the high-
frequency noise (cutoff was 300 Hz).
Survey geometry, and 3D binning
In 2000 and 2005, two sets of parallel 2D SCS lines were
acquired for 3D imaging of Bullseye Vent (Fig. 3). Each
set consists of 21 lines spaced roughly 25 m apart. The
streamer and airgun were deployed in both surveys in a
similar fashion resulting in similar parameters for cal-
culating the common mid-point locations. In both
surveys a 40 in3 (0.75 l) airgun with a wave-shape
kit that helps reducing the bubble pulse was used. The
airgun is towed at a nominal depth of 2 m below the
sea-surface. The airgun generates seismic signals over a
broad spectrum ranging from 20 to 500 Hz with a central
frequency of *120 Hz (Fig. 4). The Teledyne seismic
streamer consisted of a 25 m active section hosting 50
hydrophones each 0.5 m apart. The oil-filled streamer
generally settles at a depth of *4 m below the sea sur-
face (Riedel 2001).
The 2000 data were merged into a rectangular 3D
volume by binning the data using an inline spacing of 25 m
and a crossline spacing of 12 m. The 2005 data had
irregular shot point spacing across the survey area and also
showed significant line orientation divergence. The data
were therefore binned with a coarser bin size of 30 m in
the inline direction and 18 m in the crossline direction. The
first step in the processing sequence is re-binning of the
2005 survey and interpolation. The 2000 data served as
reference and the 2005 data were forced onto a 25 9 12 m
grid using the four nearest neighboring traces from the
2005 survey and linear interpolation.
Time- and phase-shifting
The next processing step after re-binning is a time- and
phase-matching of the 2005 data. The data sets were flat-
tened to a common datum at 1.6 s two-way travel time
(TWT) using the seafloor reflection. However, some small
vertical time offsets were still present after datuming and
need to be removed. The phase-matching is carried out by
Fig. 2 Flow chart illustrating the 4D seismic processing sequence
358 Mar Geophys Res (2007) 28:355–371
123
calculating the cross-correlation between coincident traces
from both data sets and identifying the time and phase shift
required to maximize the match between both traces over a
specifically defined window length. Since it is generally not
known where and to what magnitude changes did occur, a
horizon needs to be identified where no changes have
occurred between the two surveys. In this study the seafloor
serves as such a reference horizon. The cross-correlation
window for the time- and phase-matching was centered at
1.6 s TWT with a 10 ms running window above and below
this datum.
Gain adjustment
Both data sets were acquired with different overall gain
settings in the field. The individual 2D lines also show
amplitude variations that result in striations in time-slices.
These amplitude variations within the 2000 data were
removed prior to the 4D analyses by balancing all 2000
data relative to a reference amplitude, which was arbitrarily
chosen as the root-mean-square (RMS) amplitude of
inline 1. To remove the gain difference between the 2005
and balanced 2000 data, a gain adjustment was carried out
that calculates the RMS amplitude of all traces in the 2000
data and compares it to that of all traces in the 2005 data.
A 1 s TWT time window spanning the entire data range
(1.5–2.5 s TWT) was used to define the RMS amplitudes
and a single, global scalar is used for all traces to adjust the
2005 data.
Fig. 4 (a) Waveform and (b) frequency spectra of the original 2000
seismic data (blue), original 2005 data (red) and 2005 data with 4D
processing applied (green)
Fig. 3 Map of survey line geometry and 3D cube outline of the (a) 2000 reference data and (b) 2005 data
Mar Geophys Res (2007) 28:355–371 359
123
Shaping filter
The last step in the processing sequence is defining a
shaping filter that transforms the 2005 data to match the
2000 data for a specific horizon. The idea behind this step
is to reproduce a section out of the 2005 data that is as
close as possible to the 2000 data acquisition parameters
(source wave-form and frequency content), but which
preserves any real changes present in the sub-surface. The
flattened seafloor was again used as a reference horizon
where no changes were believed to have occurred between
the two surveys. A 20 ms long window centered on the
seafloor at 1.6 s TWT was chosen to define the shaping
filter. The result of this shaping filter is that the seafloor
reflection defined over the 20 ms long window is identical
in both surveys.
Evaluation of 4D processing results
The effect of the 4D seismic processing steps needs to be
carefully evaluated in order to avoid over-interpretation of
the differences in the 2000 and 2005 data. The first pro-
cessing step involved re-binning by trace interpolation of
the 2005 data to match survey geometries. To evaluate this
process, time-slices of the flattened 3D volumes at 1.65 and
1.68 s TWT for the reference data and the 2005 data are
compared with time-slices of the flattened data prior to
merging them into a 3D volume using the original common
mid point locations (Figs. 5 and 6). The comparison shows
that the same large-scale structural features are present in
both data prior to (Figs. 5a, b and 6a, b) and after binning
(Figs. 5c and 7c) and that the re-binning and interpolation
of the 2005 data did not introduce any false structural
information. The seismic amplitude slice at 1.65 TWT
(Fig. 5c–e) and the instantaneous amplitude or envelope
(Fig. 5f–h) clearly show that the 4D process was successful
in generating a 3D data volume out of the original 2005
data that is much closer in the overall characteristics of the
2000 data. In the envelope slices at 1.65 s TWT there are
two structural elements very well preserved in the data
after 4D processing that are very similar to those seen in
the original 2000 data: (1) Two parallel low-amplitude
lines with a high-amplitude centre are seen in the SW lower
half of the data cube. This feature is not evident in the 2005
data prior to the processing, but is again visible afterwards.
(2) In the Northeast upper corner a high-amplitude ring-
like structure is seen around a prominent blank zone.
Although this feature is recognizable in the original 2005
data, the feature is more clearly defined after applying the
4D processing sequence. The comparison of the slice at
1.65 s TWT also shows the difficulties of defining mean-
ingful differences over the centre of Bullseye Vent, where
the slice after 4D processing (Fig. 5h) shows prominent
high-amplitude spots not present in the 2000 data set.
Similar observations can be made on the time-slice at
1.68 s TWT (Fig. 6c–h). Amplitudes after 4D processing
appear more similar to the 2000 reference data and the
same 3D structures are preserved, but some significant
changes can be identified, which are described in more
detail below.
The comparison also shows that the 4D processing
sequence fails over areas with steep dips, or within areas of
accreted sediments, especially towards the southwest and
northeast corners of the 3D seismic cubes. While the ori-
ginal 2000 data show some coherent reflectivity, the
structures are not well re-produced after the 4D processing
of the 2005 data set. This is certainly partially a result of
the relative poor input data of the 2005 survey, but it also
defines the limit over which 4D time-lapse analyses should
be attempted. If the sub-bottom reflections are seafloor- to
sub-seafloor-parallel there is a high chance of succeeding
in generating meaningful data for comparison. If the
structures are too steep, the comparison becomes less
reliable. Acquiring data with higher inline and crossline
spacing or multi-fold data can help to push the limit of
structural dip that can be included in the analysis. In this
study interpretations are limited to areas with dips less than
8 � (equivalent to *2 ms TWT depth change per trace, or
*2 m depth change over a horizontal distance of 12 m
using an average velocity of 1,600 m/s in the upper 100 m
below seafloor). This limitation excludes about one-third of
the area covered by the 3D volumes, or about 300 m along
an inline on the Northeast and Southwest corner.
Evaluation of the time- and phase-matching, as well as
shaping filters, is best carried out on sections of the actual
seismic inlines. Inlines 9, 12 and 19 were chosen to
demonstrate the effects (Figs. 7–9). In the case of inline 9
the apparent blanking seen around crossline 70 in the
original 2005 data (Fig. 7b) was entirely the result of poor
airgun shots, as seen in the deteriorated seafloor reflec-
tion. The apparent blanking has been removed and
amplitudes were restored by the 4D processing. Inline 12
was chosen for this comparison because it is located over
the centre of Bullseye Vent and crosses the massive gas
hydrate cap previously identified (Riedel et al. 2002).
This cap is clearly shown in the original 2000 data
between crosslines 50 and 85 within the top 20 ms below
seafloor (Fig. 8a). The re-binned 2005 data prior to the 4D
processing do not show as strong evidence of this cap
reflection but similar subsurface blanking is apparent
(Fig. 8b). However, some of the blanking around the cap
reflector appears to be not as prominent after applying the
4D processing sequence, whereas two smaller-scale
blanking areas around crosslines 30 and 50 are still
present (Fig. 8d). This line also shows the complex nature
360 Mar Geophys Res (2007) 28:355–371
123
Fig
.5
Co
mp
aris
on
of
tim
e-sl
ices
at1
.65
sT
WT
(rel
ativ
eto
flat
ten
edse
aflo
or)
.D
ash
edg
reen
lin
ein
dic
ates
lim
ito
fd
ata\
8�
dip
.IO
DP
Sit
eU
13
28
dri
llh
ole
sar
esh
ow
nas
gre
end
ots
.
(a)
Refl
ecti
on
amp
litu
de
usi
ng
ori
gin
al2
00
0li
ne
geo
met
ry.
Am
pli
tud
esar
ed
raw
nin
a2
59
12
mlo
ng
rect
ang
ula
rce
nte
red
aro
un
dre
flec
tio
np
oin
t,(b
)ti
me-
slic
eu
sin
go
rig
inal
20
05
lin
e
geo
met
ry,
(c)
refl
ecti
on
amp
litu
de
of
bin
ned
and
gai
n-b
alan
ced
20
00
3D
cub
e,(d
)re
flec
tio
nam
pli
tud
eo
fo
rig
inal
,b
ut
re-b
inn
ed2
00
5d
ata,
(e),
refl
ecti
on
amp
litu
de
of
20
05
dat
aaf
ter
com
ple
te
4D
pro
cess
ing
,(f
)in
stan
tan
eou
sam
pli
tud
e(e
nv
elo
pe)
of
the
ori
gin
al2
00
0re
fere
nce
dat
a(b
alan
ced
),(g
)in
stan
tan
eou
sam
pli
tud
e(e
nv
elo
pe)
of
the
ori
gin
al,
bu
tre
-bin
ned
20
05
dat
a,an
d
(h)
inst
anta
neo
us
amp
litu
de
(en
vel
op
e)o
fth
e4
Dp
roce
ssed
20
05
dat
a.D
etai
lsse
ete
xt.
No
rth
arro
wis
sho
wn
in(a
),o
rig
inal
lin
eo
rien
tati
on
is3
9.6
5�.
Th
elo
cati
on
sfo
rin
lin
es9
,1
2,
and
19
as
sho
wn
inF
igs.
7–
9ar
ein
dic
ated
by
the
das
hed
yel
low
lin
es
Mar Geophys Res (2007) 28:355–371 361
123
Fig
.6
Co
mp
aris
on
of
tim
e-sl
ices
at1
.68
sT
WT
(rel
ativ
eto
flat
ten
edse
aflo
or)
.D
ash
edg
reen
lin
ein
dic
ates
lim
ito
fd
ata
\8�
dip
.IO
DP
Sit
eU
13
28
dri
llh
ole
sar
esh
ow
nas
gre
end
ots
.
(a)
Refl
ecti
on
amp
litu
de
usi
ng
ori
gin
al2
00
0li
ne
geo
met
ry.
Am
pli
tud
esar
ed
raw
nin
a2
59
12
mlo
ng
rect
ang
ula
rce
nte
red
aro
un
dre
flec
tio
np
oin
t,(b
)ti
me-
slic
eu
sin
go
rig
inal
20
05
lin
e
geo
met
ry,
(c)
refl
ecti
on
amp
litu
de
of
bin
ned
and
gai
n-b
alan
ced
20
00
3D
cub
e,(d
)re
flec
tio
nam
pli
tud
eo
fo
rig
inal
,b
ut
re-b
inn
ed2
00
5d
ata,
(e),
refl
ecti
on
amp
litu
de
of
20
05
dat
aaf
ter
com
ple
te
4D
pro
cess
ing
,(f
)in
stan
tan
eou
sam
pli
tud
e(e
nv
elo
pe)
of
the
ori
gin
al2
00
0re
fere
nce
dat
a(b
alan
ced
),(g
)in
stan
tan
eou
sam
pli
tud
e(e
nv
elo
pe)
of
the
ori
gin
al,
bu
tre
-bin
ned
20
05
dat
a,an
d
(h)
inst
anta
neo
us
amp
litu
de
(en
vel
op
e)o
fth
e4
Dp
roce
ssed
20
05
dat
a.D
etai
lsse
ete
xt.
No
rth
arro
wis
sho
wn
in(a
),o
rig
inal
lin
eo
rien
tati
on
is3
9.6
5�.
Th
elo
cati
on
sfo
rin
lin
es9
,1
2,
and
19
as
sho
wn
inF
igs.
7–
9ar
ein
dic
ated
by
the
das
hed
yel
low
lin
es
362 Mar Geophys Res (2007) 28:355–371
123
of the centre of the cold vent around the massive gas
hydrate cap. The artificial vertical shifts (specifically at
crossline 78 and 82) in the 2005 data after 4D processing
are a result of the irregular seafloor reflection around the
cap and the chosen window length for the cross-correla-
tion to define the shaping-filter. Any changes observed on
time-slices around the centre of the blank zone should
therefore be treated cautiously.
Fig. 7 Comparison of seafloor-flattened seismic sections of inline 9
from Bullseye Vent data sets. Shown are seismic amplitudes. (a)
Reference data from 2000, (b) re-binned 2005 data (no processing
applied), (c) re-binned data of 2005 with time- and phase-shifting and
overall gain adjustment applied, (d) re-binned 2005 data with
complete 4D processing sequence applied
Mar Geophys Res (2007) 28:355–371 363
123
Inline 19 showed highly variable seafloor waveforms
between crosslines 10 and 50 in the original 2005 data
(Fig. 9b) and apparent vertical subsurface blanking.
However, after complete 4D processing, this artificial
blanking was removed (Fig. 9d). Further along the same
line, between crosslines 70 and 90, the original 2005 data
showed wide-spread subsurface blanking, which is not
vertically constrained compared to the blanking seen
Fig. 8 Comparison of seafloor-flattened seismic sections of inline 12
over centre of Bullseye Vent with gas hydrate cap-reflector. Shown
are seismic amplitudes (negative amplitude in red, positive amplitude
in black). (a) Reference data from 2000, (b) re-binned 2005 data (no
processing applied), (c) re-binned data of 2005 with time- and phase-
shifting and overall gain adjustment applied, (d) re-binned 2005 data
with complete 4D processing sequence applied
364 Mar Geophys Res (2007) 28:355–371
123
between crosslines 10 and 50 (Fig. 9b). This blanking
was preserved by the 4D processing (Fig. 9d) and is not
seen in the 2000 reference data (Fig. 8a). It is therefore
interpreted as a real geologic signal and evidence for
temporal changes in the subsurface blanking of Bullseye
Vent.
Fig. 9 Comparison of seafloor-flattened seismic sections of inline 19
from Bullseye Vent data sets. Shown are seismic amplitudes (negative
amplitude in red, positive amplitude in black). (a) Reference
data from 2000, (b) re-binned 2005 data (no processing applied),
(c) re-binned data of 2005 with time- and phase-shifting and overall
gain adjustment applied, (d) re-binned 2005 data with complete 4D
processing sequence applied. Details see text
Mar Geophys Res (2007) 28:355–371 365
123
Mapping changes in subsurface blanking using seismic
attributes
The aim of this section is to demonstrate what seismic
attributes are best suited for showing differences in the sub-
surface properties, as well as to show that significant changes
did occur over the time span between the two surveys.
Seismic blanking is linked to the presence of gas hydrate and/
or free gas, as outlined above. Changes in the area affected by
blanking indicate that the amount and/or location of gas
hydrate (and/or free gas) have changed. Mapping these
changes accurately is the first step in developing a geologic
scenario that may gave rise to those changes.
Seismic blanking was previously identified for the cold
vents on the northern Cascadia margin and mapping using
time-slices of instantaneous amplitude showed character-
istic high-amplitude rims around the blank zones for all
four main cold vents identified (Riedel 2001; Riedel et al.
2002; Wood et al. 2000). Those amplitude rims were
interpreted as the result of constructive interference of
diffractions from the top of the gas hydrate cap with dee-
per, regular reflectivity (Riedel 2001; Riedel et al. 2002).
In the case of Bullseye Vent, a prominent ring structure
was identified surrounding the central blank zone. The
central blank zone is outlined in all time-slices shown in
Figs. 5 and 6. Blanking that is easily identified on vertical
seismic sections displaying regular seismic amplitude with
phase information preserved, is more difficult to follow on
time-slices showing amplitudes of arbitrary phases (see
e.g., Figs. 5a–c and 6a–c). Removing phase information by
calculating the seismic envelope (instantaneous amplitude)
shows a much clearer image of the area affected by
blanking (Figs. 5f–h and 6f–h).
The geometrical attribute ‘‘similarity’’ or ‘‘coherence’’
(Taner 2000 and references therein) appears to generate
clearer images of blanking and is thus used in mapping
those areas. This attribute is computed over a specific
window size, which is larger than the length of the domi-
nant wavelength of the data. It identifies the overall
similarity of a trace and its nearest neighbors. For the
Bullseye Vent data the similarity attribute is computed over
a frequency range from 20 to 160 Hz, a window length of
10 ms and incorporates the nearest four neighboring traces.
The time-slice of the similarity attribute at 1.65 s TWT
of the reference 2000 data shows several clear areas of
reduced seismic similarity (Fig. 10a) that are not as easily
Fig. 10 Time-slices of similarity attribute 1.65 s TWT using (a)
2000 reference data (balanced), (b) 2005 data with 4D processing
applied. Darker colors represent higher similarity. (c) Time-slice of
difference between similarity shown in (a) and (b) of the two surveys.
The 2005 data were subtracted from the 2000 data. Red colors show
positive difference (enhanced blanking, less similarity in 2005), black
colors are negative difference (reduced blanking, higher similarity in
2005), (d) line drawing outlining main features (labeled 1 and 2)
compared in study; for details see text. North arrow is shown in (d).
The color scales for the similarity attributes in (a) and (b) are between
0 (white) and 1.5 (black); the color scale for the difference plot in (c)
is between -0.225 (black) and +0.225 (red). IODP Site U1328 drill
holes are shown as black dots. The locations for inlines 9, 12, and 19
as shown in Figs. 7–9 are indicated by the dashed yellow lines
366 Mar Geophys Res (2007) 28:355–371
123
identified from the slice of instantaneous amplitude
(Fig. 5f). The centre of Bullseye Vent covers an area of
400 m (in the Southwest–Northeast direction) by 270 m (in
the SE–NW orientation). Towards the northeast two addi-
tional smaller blank areas surrounded by rims of high
similarity are identified. The traces of two low-similarity
lineaments earlier identified on time-slices of instantaneous
amplitude are also clearly visible in the lower southwestern
half of the similarity time-slice. The time-slice at the same
depth generated from the 2005 data with 4D processing
applied shows an overall similar structure. A strikingly
similar element is the pair of low-similarity traces in the
lower southwestern half of the cube.
The two smaller-sized blank areas to the northeast of the
centre of Bullseye Vent are preserved, although the mag-
nitude of similarity has changed compared to 2000. This is
best visualized on a time-slice of the difference in simi-
larity (Fig. 10c). The color code used represents positive
differences in red and negative difference in black. Since
the difference was calculated by subtracting the similarity
of the 2005 processed data from the 2000 reference data,
red areas identify zones where similarity is lower in 2005
than in 2000, i.e., the traces show less coherent reflectivity
in 2005 than in 2000. Similarly areas in black identify
zones where similarity in the 2005 data is larger than in
2000 reference. The red areas are zones where blanking
intensified since 2000 and black areas are zones of reduced
blanking.
On the time-slice at 1.65 s TWT the area labeled ‘‘1’’ in
the Northeast corner is a blank area that has ‘‘healed’’ since
2005 as blanking appears to be reduced (Fig. 10d). This
can also been seen on inline 12 (Fig. 8a, d). The centre of
Bullseye Vent is dominated by prominent red colors
(labeled ‘‘2’’) in the difference plot (Fig. 10c, d) showing
an overall increase in blanking. The time-slice of similarity
difference at 1.68 s TWT shows almost entirely red colors
(labeled ‘‘2’’) over the centre and surrounding areas of
Bullseye Vent (Fig. 11). Blanking appears to have not only
been intensified, but also spread to a larger area. The same
small blank zone labeled ‘‘1’’ that showed apparent healing
at shallower depth is completely absent at this depth in the
2005 data, whereas it can still be identified in the 2000
data. An area (East–West trending) with an apparent
increase in similarity dominates the southwestern part of
Fig. 11 Time-slices of similarity attribute 1.68 s TWT using (a)
2000 reference data (balanced), (b) 2005 data with 4D processing
applied. Darker colors represent higher similarity. (c) Difference
between similarity shown in (a) and (b) of the two surveys. The 2005
data were subtracted from the 2000 data. Red colors show positive
difference (enhanced blanking, less similarity in 2005), black colors
are negative difference (reduced blanking, higher similarity in 2005),
(d) line drawing outlining main features (labeled 1, 2 and 3)
compared in study; for details see text. North arrow is shown in (d).
The color scales for the similarity attributes in (a) and (b) are between
0 (white) and 1.5 (black); the color scale for the difference plot in (c)
is between -0.225 (black) and +0.225 (red). IODP Site U1328 drill
holes are shown as black dots. The locations for inlines 9, 12, and 19
as shown in Figs. 7–9 are indicated by the dashed yellow lines
Mar Geophys Res (2007) 28:355–371 367
123
Bullseye Vent at this depth (labeled ‘‘3’’ (Fig. 11); how-
ever, this change in similarity is not as prominent as other
changes identified between the surveys.
Interpretation of changes in similarity on individual
time-slices may be confusing and not inherently convinc-
ing. However, when zones of increased seismic similarity
identified on Figs. 10 and 11 are traced from time-slice to
time-slice, they form downward continuations, resembling
somewhat the nature of fracture zones (Fig. 12). In Fig. 12
the similarity of the 2000 data along inline 17 is compared
to the similarity after 4D processing. The difference plot
highlights the changes in similarity (red colors indicate
increased blanking in 2005, i.e., reduced similarity). Most
of the changes resemble the form of semi-vertical fracture
zones. There are also two zones showing horizontally
extended, or ‘‘bedded’’-like, areas of reduced similarity in
2005 between crossline 50 and 80, at 1.68 and 1.78 s TWT,
respectively. It should be emphasized that this interpreta-
tion is still somewhat speculative in nature and needs
verification by additional 4D seismic imaging.
Discussion and significance of observed changes
Although changes in the seismic amplitudes and related
attributes (especially similarity) of the two data sets were
identified, significant uncertainties in their meaning
remain. Are these meaningful results or artifacts of the 4D
processing sequence?
The seismic processing sequence applied tried to
remove differences that result from the acquisition induced
variations of source signature and frequency content. The
re-binning of the 2005 data to match the same 3D geometry
of the 2000 data has not introduced any false structural
elements and is believed to be adequate for the purpose of
comparing the two data sets with the constraint of dips of
less than 8�. Using the similarity attribute to define areas
affected by blanking is more robust than the actual
amplitude itself because it removes phase information. The
similarity attribute was further calculated over a 10 ms
long window and frequencies were limited to 160 Hz,
which results in an overall smoother image, leaving only
larger-scale features within the data. Thus the process is
less affected by high-frequency noise.
The 4D processing sequence relies on the assumption
that the seafloor is unchanged between the two surveys and
that the trace-by-trace shaping filter can be applied and
does not introduce artifacts. However, if changes did occur
in the near-seafloor range (e.g., within the centre of the
vent) covered by the filter length, then those changes would
be propagated into the subsurface and subsequently inter-
preted as deep-rooted fluid-flow related changes. To
overcome this problem, only seismic data away from the
centre and gas hydrate cap could be utilized to create
global shaping filter parameters; however, the data sets
used in this study cover too small an area to be significantly
distant from the centre of Bullseye Vent. Thus there is a
remaining uncertainty for the origin of apparent changes
below the centre of the vent and any mapped changes
should be treated with caution.
If the observed changes are believed to be real, what
caused the change in blanking? Several differing models
were brought forward to explain the nature of Bullseye
Vent and its associated seismic blanking (Riedel et al. 2002,
2006a; Wood et al. 2002; Zuhlsdorff and Spiess 2004).
As described above, the model by Riedel et al. (2006a)
Fig. 12 Comparison of seismic similarity attribute of inline 17. (a)
Similarity from the 2000 data set, (b) similarity from 2005 data after
4D processing applied, (c) difference in similarity showing apparent
new fractures zones and semi-horizontal beds of reduced similarity in
2005. The color scales for the similarity attributes in (a) and (b) are
between 0 (white) and 1.5 (black); the color scale for the difference
plot in (c) is between -0.225 (black) and +0.225 (red). The locations
for inlines 9, 12, and 19 as shown in Figs. 7–9 are indicated by the
dashed yellow lines
368 Mar Geophys Res (2007) 28:355–371
123
describes Bullseye Vent as a network of gas hydrate-filled
fractures with a cap of massive gas hydrate in the top 40–
50 mbsf. Small-scale vent outlets of only a few square
meters in size were observed at the seafloor where tempo-
rarily methane gas can escape through the gas hydrate
stability zone into the ocean. As gas migrates upward
through the network of fractures, it partially transforms into
gas hydrate, potentially blocking some of the fractures and
sealing them for further gas transport. The naturally buoy-
ant gas, which cannot combine with surrounding water to
form gas hydrate, needs to find new pathways, potentially
opening new fractures. Additional gas hydrate is likely to be
formed along those new pathways. Thus, blanking as a
result of scattering at fractures and potentially from the
presence of free gas can change over time. Although most
of the changes in similarity resemble semi-vertical fracture
paths, several zones of reduced similarity in 2005 were seen
that are ‘‘bedded’’-like or semi-horizontal. Those horizontal
extents of reduced similarity could be related to horizontal
migration of methane gas along more porous (sandy)
turbidite layers, attracting gas hydrate formation relative to
the surrounding mud. Several of those turbidite layers filled
with gas hydrate embedded in gas hydrate free mud were
recovered during IODP Expedition 311 (Riedel et al.
2006b). Increased blanking from those hydrate-filled
turbidites can be explained by the model of Lee and Dillon
(2001), where the seismic reflectivity between turbidite
layers and regular mud-dominated sequences is reduced
by the preferential accumulation of gas hydrate in the tur-
bidite layer.
It can further be speculated that gas migration can be
initiated and intensified by earthquake activity. Gas accu-
mulates below the gas hydrate stability zone at the bottom
simulating reflector, or within the cold vent where it is
sealed off from free water and could be rapidly released as
earthquake shaking, helps to open fractures and generate
new pathways.
Future opportunities
As part of the NEPTUNE program it is planned to place
broad-band seismometers near Bullseye Vent recording
earthquake activity. This will be complemented with sea-
floor-based video observations and long-time geochemical
monitoring of the fluid flux. For the second phase of the
IODP Expedition 311 it is proposed to install special
borehole monitoring instruments, including fiber-optic
temperature sensors and fluid samplers. Ideally a set of
permanent seismic receivers should be placed across
Bullseye Vent, allowing for simple and fast acquisition of
4D time-lapse seismic data. With receivers permanently
implemented on the seafloor many of the above described
uncertainties in the 4D time-lapse results could be
removed. The seismic source should also be towed at
greater water depth to remove source-signature variations
from weather induced sea surface conditions. However, the
source should be placed at such a depth that interference
between the primary reflections and the sea-surface ghost
are avoided.
Summary and conclusion
Two 3D single channel seismic reflection data sets were
used for a 4D time-lapse analysis of an active cold vent
(Bullseye Vent). The required processing steps are descri-
bed and imaging techniques are introduced to help best
identify areas of seismic blanking. The data set acquired in
2000 has superior navigation accuracy and serves as a
reference in the applied processing sequence. The 2005
data was re-binned to achieve identical 3D geometries for
the data sets and was subsequently processed using time-
and phase-matching, amplitude adjustment and shape-
filtering. The phase- and shape filters were generated using
the seafloor reflection as reference horizon where no
changes were expected. All seismic data were also flattened
using the seafloor reflection to a common datum (1.6 s
TWT). The 4D processing sequence yielded a data set from
the 2005 data that is most comparable to the conditions
under which the 2000 data were acquired. The area of
blanking (indicative of the presence of gas hydrate and/or
free gas) was defined using seismic attributes such as
instantaneous amplitude and similarity. The seismic simi-
larity attribute of the 4D processed data was subtracted
from the reference similarity volume and was used to
identify areas of apparent change.
The centre of Bullseye Vent and an area around the
centre was seen to be characterized by intensified blanking.
Tracing changes from time-slice to time-slice allowed the
definition of new pathways that are interpreted as newly
formed fractures/faults that allow upward migration of
fluids. Bullseye Vent has previously been characterized by
a subsurface network of fractures that are partially gas
hydrate filled feeding methane gas to shallow depths. Gas
hydrate is also occurring preferentially in coarser-grained
turbidite layers. Seismic reflectivity between turbidite
layers and regular mud-dominated sequences can be
reduced by the presence of gas hydrate in the turbidite
layer. Thus blanking reflects the presence of gas hydrate
and/or free gas in either fractures or sandy layers. Vent
outlets with chemosynthetic communities and periodic gas
escapes that were identified by bottom-video observations
are the seafloor expression of such complicated subsurface
networks. Upward-migrating gas (potentially intensified by
earthquake activity) can open new fracture pathways,
Mar Geophys Res (2007) 28:355–371 369
123
resulting in a switch of the active vent outlet at the surface
and also a change of the area affected by blanking in the
subsurface. The mapped changes between the 2000 and
2005 data set may indicate that new pathways for upward-
migrating methane gas were generated and that gas hydrate
was newly formed in areas with increased blanking. Areas
where blanking was reduced between the two surveys may
reflect areas where formerly trapped free gas in fractures
(disconnected to surrounding water to form gas hydrate)
may have been liberated and moved along new fractures
resulting in the overall ‘‘healing’’ of the blank area.
The results of this 4D time-lapse imaging showed that
even relatively low-quality 3D SCS data can be used to
detect subsurface changes if the seismic data is processed
appropriately.
Acknowledgements The author would like to acknowledge the
important contributions of the Coast Guard crews onboard the
research vessel John P. Tully and scientists involved in the data
acquisition of the two data sets, especially George Spence and Ele
Willoughby. Furthermore the author wants to acknowledge Seismic
Micro Technology for the use of Kingdom Suite and Hampson &
Russell for the use of the program PRO4D used in this analysis.
Additional thanks go to Gilles Bellefleur, Mathieu Duchesne, and Ele
Willoughby for many helpful suggestions, discussions and encour-
agements to carry out this study.
References
Beaudet F, Riedel M, Chapman NR (2001) ROPOS gas hydrates 2001:
a seafloor survey at methane cold seeps offshore Vancouver
Island. CEOR report 2001–2002, University of Victoria, Canada,
April 30–May 7, 2001
Collier R, Klinkhammer G, Torres M, Trehu A, Heeschen K, Rehder
G, Suess E, de Angelis M, Carnocki H, Whiticar M, Barrazoul L,
Eby P, Eek M, Grant N, Schafer H, Nakamura K (1999) Methane
distributions and fluxes in the water column above an emerging
methane hydrate field on the Cascadia Accretionary Prism. Fall
AGU, San Francisco, December 13–17, 1999
Eastwood JE, Johnston D, Huang X, Craft K, Workman R (1998)
Processing for robust time-lapse seismic analysis: Gulf of
Mexico example, Lena field. In: 68th ann. internat. mtg. soc.
expl. Geophys., Expanded Abstracts, pp 20–23
Gan L, Yao F, Hu Y, Liu Y, and Du W (2004) Applying 4D seismic
to monitoring water drive reservoir. SEG expanded abstracts
23:2553. doi:10.1190/1.1839705
Harris PE, Henry B (1998) Time-lapse processing: a north sea case
study. In: 68th ann. internat. mtg. soc. expl. Geophys., Expanded
Abstracts, pp 1–4
Hobro JWD, Minshull TA, Singh SC, Chand S (2005) A three-
dimensional seismic tomographic study of the gas hydrate stability
zone, offshore Vancouver Island. J Geophys Res 110:B09102.
doi:10.1029/2004JB003477
Kobayashi K (2002) Tectonic significance of the cold seepage zones
in the eastern Nankai accretionary wedge—an outcome of the
15 years’ KAIKO projects. Mar Geol 187:3–30
Lee MW, Dillon WP (2001) Amplitude blanking related to the pore-
filling of gas hydrate in sediments. Mar Geophys Res 22:101–109
Lumley DE (2001) Time-lapse seismic reservoir monitoring. Geo-
physics 66(1):50–53
Naess OE (2006) Repeatability and 4D seismic acquisition. SEG
expanded abstracts 25:3300. doi:10.1190/1.2370217
Novosel I (2002) Physical properties of gas hydrate related sediments
offshore Vancouver Island. M.S. thesis, University of Victoria,
Victoria, 10 December, 2002, 114 pp
Novosel I, Spence GD, Hyndman RD (2005) Reduced magnetization
produced by increased methane flux at a gas hydrate vent. Mar
Geol 216:265–274
Paull CK, Matsumoto R, Wallace P (1996) Proceedings of the ocean
drilling program, initial reports 164. Ocean Drilling Program,
College Station 623 pp
Pohlman J, Spence GD, Chapman NR, Hyndman RD, Grabowski KS,
Coffin RB (2003) Evidence for anaerobic methane oxidation in
gas hydrate rich sediments on the northern Cascadia margin
offshore Vancouver Island. EGS-AGU-EUG Joint Assembly,
Nice-France, April 6–11
Riedel M (2001) 3D seismic investigations of northern Cascadia
marine gas hydrates. PhD thesis, University of Victoria, Victoria,
14 September, 305 pp
Riedel M, Spence GD, Chapman NR, Hyndman RD (2002) Seismic
Investigations of a vent field associated with gas hydrates,
Offshore Vancouver Island. J Geophys Res JGR Solid Earth
107(B9):2200. doi:10.1029/2001JB000269
Riedel M, Novosel I, Spence GD, Hyndman RD, Chapman NR,
Solem RC, Lewis T (2006a) Geophysical and geochemical
signatures associated with gas hydrate related venting at the
north Cascadia margin. GSA Bull 118(1/2). doi:10.1130/
B25720.1
Riedel M, Collett TS, Malone MJ, Expedition 311 Scientists (2006b)
Proceedings of IODP, vol 311. Integrated Ocean Drilling
Program Management International, Inc., Washington. doi:
10.2204/iodp.proc.311.2006
Sassen R, Sweet ST, Milkov AV, DeFreitas DA, Kennicutt II MC
(2001) Stability of thermogenic gas hydrate in the Gulf of
Mexico: constraints on models of climate change. In: Paull CK,
Dillon WP (eds) Natural gas hydrates: occurrence, distribution,
and detection, vol 124. Am. Geophysical Union, Geophys
Monogr Ser, pp 131–144
Schwalenberg K, Willoughby EC, Mir R, Edwards RN (2005) Marine
gas hydrate signatures in Cascadia and their correlation with
seismic blank zones. First Break 23:57–63
Shimeld J, Mosher D, Louden K, LeBlanc C, Osadetz K (2004)
Bottom simulating reflectors and hydrate occurrences beneath
the Scotian slope offshore Eastern Canada. In: AAPG Hedberg
conference 2004 ‘‘Gas hydrates: energy resource potential and
associated geologic hazards’’ Vancouver, BC, Canada, Septem-
ber 12–16, 2004
Solem RC, Spence GD, Vukajlovich D, Hyndman RD, Riedel M,
Novosel I, Kastner M (2002) Methane advection and gas hydrate
formation within an active vent field offshore Vancouver Island.
In: Proceedings of the 4th international conference on gas
hydrate, Yokohama
Spence GD, Minshull TA, Fink C (1995) Seismic structure of
methane gas hydrate, offshore Vancouver Island. Proc Ocean
Drill Program Sci Res 146:163–174
Suess E, Bohrmann G, von Huene R, Linke P, Wallmann K, Lammers
S, Sahling H (1998) Fluid venting in the eastern Aleutian
subduction zone. J Geophys Res 103:2597–2614
Suess E, Torres ME, Bohrmann G, Collier RW, Greinert J, Linke P,
Rehder G, Trehu A, Wallmann K, Winckler G, Zuleger E (1999)
Gas hydrate destabilization: enhanced dewatering, benthic
material turnover and large methane plumes at the Cascadia
convergent margin. Earth Planet Sci Lett 170:1–15
Suess E, Torres ME, Bohrmann G, Collier RW, Rickert D, Goldfinger
C, Linke P, Heuser A, Sahling H, Heeschen K, Jung C, Nakamura
K, Greinert J, Pfannkuche O, Trehu AM, Klinkhammer G,
370 Mar Geophys Res (2007) 28:355–371
123
Whiticar MJ, Eisenhauer A, Teichert B, Elvert M (2001) Sea floor
methane hydrates at hydrate ridge: Cascadia margin. In: Paull CK,
Dillon WP (eds) Natural gas hydrates: occurrence, distribution,
and detection, vol 124. Am. Geophysical Union, Geophys Monogr
Ser, pp 87–98
Taner MT (2000) Attributes revisited. http://www.rocksolidimages.
com/pdf/attrib_revisited.htm. Cited August 9, 2005
Tyron MD, Brown KM, Torres ME, Trehu AM, McManus J, Collier
RW (1999) Measurements of transience and downward fluid
flow near episodic methane gas vents, Hydrate Ridge, Cascadia.
Geology 27(12):1075–1078
von Rad U, Berner U, Delisle G, Doose-Rolinski H, Fechner N, Linke P,
Luckge A, Roeser HA, Schmaljohann R, Wiedicke M, SONNE 122/
130 Scientific Parties (2000) Gas and fluid venting at the Makran
Accretionary Wedge off Pakistan. Geo-Mar Lett 20:10–19
Willoughby EC, Schwalenberg K, Edwards RN, Spence GD,
Hyndman RD (2005) Assessment of marine gas hydrate
deposits: a comparative study of seismic, electromagnetic and
seafloor compliance methods. In: International conference on gas
hydrates, Trondheim, Norway
Wood WT, Lindwall DA, Gettrust JF, Sekharan KK, Golden B (2000)
Constraints on gas or gas hydrate related wipeouts in seismic
data through the use of physical models. Eos Trans Am Geophys
Union 81(48):F639
Wood WT, Gettrust JF, Chapman NR, Spence GD, Hyndman RD
(2002) Decreased stability of methane hydrates in marine
sediments owing to phase-boundary roughness. Nature 420:
656–660
Zuhlsdorff L, Spiess V (2004) Three-dimensional seismic character-
ization of a venting site reveals compelling indications of natural
hydraulic fracturing. Geology 32(2):101–104
Zykov MM, Chapman NR (2004) 3-D velocity model of hydrocarbon
vent site in Cascadia region offshore Vancouver Island. In:
AAPG Hedberg conference 2004 ‘‘Gas hydrates: energy resource
potential and associated geologic hazards’’, Vancouver, BC,
Canada, September 12–16, 2004
Mar Geophys Res (2007) 28:355–371 371
123